Wan 2.2 S2V 14B (speech to video) on AMD Radeon RX 7900 XTX

Can the AMD Radeon RX 7900 XTX run Wan 2.2 S2V 14B (speech to video)?

75 pts

Yes, with little memory to spare

Tight fit. At Q4 GGUF it needs about 21.6 GB of 24.0 GB usable, with an estimated speed of 22.3 min.

Best precision: Q4 GGUF
Memory mode: Everything on GPU
Time per clip: 22.3 min
Output: 832×480, 81 frames, 40 steps

Download Wan 2.2 S2V 14B (speech to video)

Recommended for the AMD Radeon RX 7900 XTX: q4_gguf. Every other precision has its own download link in the table below.

Quantized files · GGUF for ComfyUI

Smaller GGUF versions for ComfyUI (with the ComfyUI-GGUF node). No sign-in needed; the model’s license still applies.

Download the q4_gguf (Q4_K_S) file (opens in a new tab)

huggingface.co/QuantStack/Wan2.2-S2V-14B-GGUF (opens in a new tab)

Original weights · Hugging Face

Full-precision files from Alibaba, for fine-tuning, vLLM, Transformers, Diffusers or ComfyUI.

huggingface.co/Wan-AI/Wan2.2-S2V-14B (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Alibaba. The weights themselves download from Hugging Face.

github.com/Wan-Video/Wan2.2 (opens in a new tab)

git clone https://github.com/Wan-Video/Wan2.2

Links checked 5 days ago against Hugging Face, Ollama and GitHub.

How the A grade is scored

Every result gets up to 100 points from four parts. The points decide the letter grade.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	31.8 / 45
Memory fitFull marks when everything sits in GPU or unified memory; fewer when part spills to system RAM.	23.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	5.5 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	15.0 / 15
Total	75.3 / 100	75 ptsGreat (A) · 75 of 100 points: Model quality 31.8/45, Memory fit 23.0/25, Speed 5.5/15, Memory use 15.0/15

Grade scale: S 85+ · A 72+ · B 58+ · C 42+ · D 25+ · F under 25 points.

Every precision on the AMD Radeon RX 7900 XTX

Precision	Memory needed	Available	Time per output	Memory mode	Fit	Grade · points	Download
`FP16`	17.0 GB	24.0 GB	24.1 min	Sequential CPU offload	Runs well	66 ptsGood (B) · 66 of 100 points: Model quality 34.2/45, Memory fit 12.0/25, Speed 5.3/15, Memory use 15.0/15	Original weights (opens in a new tab)
`FP8`	22.1 GB	24.0 GB	17.5 min	Text encoder offloaded	Tight fit	72 ptsGreat (A) · 72 of 100 points: Model quality 33.5/45, Memory fit 18.0/25, Speed 6.2/15, Memory use 15.0/15	Original weights (opens in a new tab)
`Q8 GGUF`	11.7 GB	24.0 GB	27.8 min	Sequential CPU offload	Runs well	62 ptsGood (B) · 62 of 100 points: Model quality 33.9/45, Memory fit 12.0/25, Speed 4.9/15, Memory use 11.2/15	GGUF (opens in a new tab)
`Q4 GGUF`	21.6 GB	24.0 GB	22.3 min	Everything on GPU	Tight fit	75 ptsGreat (A) · 75 of 100 points: Model quality 31.8/45, Memory fit 23.0/25, Speed 5.5/15, Memory use 15.0/15	GGUF (opens in a new tab)

Other hardware that runs it

78 ptsGreat (A) · 78 of 100 points: Model quality 33.5/45, Memory fit 23.0/25, Speed 7.3/15, Memory use 15.0/15 NVIDIA GeForce RTX 5090 FP8 · 12.5 min
78 ptsGreat (A) · 78 of 100 points: Model quality 34.2/45, Memory fit 25.0/25, Speed 7.4/15, Memory use 12.1/15 NVIDIA RTX PRO 6000 Blackwell FP16 · 12.2 min
75 ptsGreat (A) · 75 of 100 points: Model quality 34.2/45, Memory fit 25.0/25, Speed 4.6/15, Memory use 12.1/15 AMD Ryzen AI Max+ 395 (128 GB) FP16 · 32.0 min
75 ptsGreat (A) · 75 of 100 points: Model quality 33.9/45, Memory fit 25.0/25, Speed 1.9/15, Memory use 15.0/15 Apple M4 Pro (48 GB) Q8 GGUF · 3.1 h
75 ptsGreat (A) · 75 of 100 points: Model quality 31.8/45, Memory fit 23.0/25, Speed 5.8/15, Memory use 15.0/15 NVIDIA GeForce RTX 4090 Q4 GGUF · 20.0 min
75 ptsGreat (A) · 75 of 100 points: Model quality 33.9/45, Memory fit 25.0/25, Speed 2.5/15, Memory use 14.1/15 Apple M4 Max (64 GB) Q8 GGUF · 1.9 h

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.